Data Preparation

April 11, 2023 · View on GitHub

Data Preparation

Overall Structure

└── RoboDepth 
      │── kitti_data
      │    │── 2011_09_26
      │    │── ...
      │    │── kitti_c
      │    └── val
      │── cityscapes
      │    ├── camera
      │    │   ├── train
      │    │   └── val
      │    ├── disparity_trainvaltest
      │    │   └── disparity
      │    ├── leftImg8bit_trainvaltest
      │    │   └── leftImg8bit
      │    └── split_file.txt
      │── nyu
      │    │── basement_0001a
      │    │── basement_0001b
      │    │── ...
      │    │── nyu_c
      │    └── split_file.txt
      └── ...

Outline

KITTI

You can download the entire raw KITTI dataset by running:

wget -i splits/kitti_archives_to_download.txt -P kitti_data/

Then unzip with:

cd kitti_data/
unzip "*.zip"
cd ..

:dart: This dataset weighs about 175GB, so make sure you have enough space to unzip too!

The train/test/validation splits are defined in the splits/ folder. By default, the code will train a depth estimation model using Zhou's subset of the standard Eigen split of KITTI, which is designed for monocular training. You can also train a model using the new benchmark split or the odometry split by setting the --split flag.

KITTI-C

The corrupted KITTI test sets under Eigen split can be downloaded from Google Drive with this link.

Alternatively, you can directly download them to the server by running:

cd kitti_data/

wget --load-cookies /tmp/cookies.txt "https://docs.google.com/uc?export=download&confirm=$(wget --quiet --save-cookies /tmp/cookies.txt --keep-session-cookies --no-check-certificate 'https://docs.google.com/uc?export=download&id=1Ohyh8CN0ZS7gc_9l4cIwX4j97rIRwADa' -O- | sed -rn 's/.*confirm=([0-9A-Za-z_]+).*/\1\n/p')&id=1Ohyh8CN0ZS7gc_9l4cIwX4j97rIRwADa" -O kitti_c.zip && rm -rf /tmp/cookies.txt

Then unzip with:

unzip kitti_c.zip

:dart: This dataset weighs about 12GB, make sure you have enough space to unzip too!

Cityscapes

Coming soon.

Cityscapes-C

Coming soon.

NYUDepth2

You can download the NYU Depth Dataset V2 from Google Drive with this link.

Alternatively, you can directly download it to the server by running:

wget --load-cookies /tmp/cookies.txt "https://docs.google.com/uc?export=download&confirm=$(wget --quiet --save-cookies /tmp/cookies.txt --keep-session-cookies --no-check-certificate 'https://docs.google.com/uc?export=download&id=1wC-io-14RCIL4XTUrQLk6lBqU2AexLVp' -O- | sed -rn 's/.*confirm=([0-9A-Za-z_]+).*/\1\n/p')&id=1wC-io-14RCIL4XTUrQLk6lBqU2AexLVp" -O nyu.zip && rm -rf /tmp/cookies.txt

Then unzip with:

unzip nyu.zip

:dart: This dataset weighs about 6.2GB, which includes 24231 image-depth pairs as the training set and the standard 654 images as the validation set.

NYUDepth2-C

Coming soon.